A Domain-Specific Language Framework for Specification and Generalization of Robot Motion

This paper presents a novel framework for trajectory specification and generation in robotic systems combining a Domain-Specific Language (DSL) with a Neural Network (NN) model. The DSL allows users to intuitively define robot motions using high-level commands, abstracting away the complexities of low-level control parameters. The NN model learns from trajectories created using Dynamic Movement Primitives (DMPs) to generate smooth and accurate robot motions. We demonstrate the effectiveness of our approach with experiments on a robotic arm platform, showcasing the framework's ability to be used in a real-world scenario. Finally, we discuss the potential applications and future directions for enhancing the framework, including the integration of advanced features into the DSL, human-robot interaction enhancements, and cognitive evaluation of the DSL interface.